F0 Dev. Timeline
F0 Inception
I was using remove.bg as a normal lazy user who couldnt help it, its slow, its expensive, its shitty. Then, Theo dropped picthing. I noticed its super fast, and I have $6 on me but totally worht it specially with the optimized image hosting .engineering thing
Research and Planning
There are a lot of bg removal services out there, some are open source and some are not, some are in-browser processing and some server-side, some use machine learning and some do not. Dove deep into various image segmentation algorithms and machine learning models, and ended up going for rembg, built upon an ONNX runtime-based gpu optimized version. Then, I build a small endpoint in express with bun runtime for serving and processing, It is not open source, neither and avaialable API to call/use yourself (yet).
F0 Initiation
I have been using and obsessing about V0. with shadcn guy, the guy who made it too, working at Vercel and all whats happening is just so inspiring for me because I love those bad boys so much they DO good for the web. So F0 came really inspired by all of this (logo, brand name, and entity)
F0.0.0.1 Out
First Commit, It came out as nexjs app with tailwind (ofcourse), drizzle ORM with neonDB (first time trying this serverless) with postgreSQL. This stack was not bad at the moment and still. It has been incredible dev experience with typeScript tho.Go open source
Beta Launch
Released a beta version to 500 selected users. Implemented A/B testing for different pricing models. Collected and analyzed user feedback, achieving a satisfaction rate of 88%.
Performance Optimization
Based on beta feedback, optimized the service for faster processing times. Implemented a caching system to reduce load times by 60%. Improved the algorithm's accuracy to 95% on standard test sets.
Official Launch
Launched the background removal service to the public. Implemented a freemium model with tiered pricing. Achieved 10,000 sign-ups in the first week with a 5% conversion rate to paid plans.